统计实验设计与科技创新

Statistics Test Design and Innovation of Science and Technology

10.6338/JDA.200704_2(2).0006

张德存(De-Cun Zhang)

科技创新 ； 正交试验设计 ； 正交表 ； innovation of science and technology ； orthogonal test design ； orthogonal table

Journal of Data Analysis

2卷2期（2007 / 04 / 01）

137 - 147

簡體中文

随着科学技术的发展，许多科研课题都要求由静态（单因素）研究转向动态（多因素）研究，即在许多重要因素同时变化的条件下，来研究事物的变化规律。那么，怎样选取以及如何分析试验结果，才能科学的回答如下问题：（1）各因素对指标的影响，哪个因素重要？哪个因素次之？（2）对于某个因素，哪个水平为好？（3）各个因素和水平依哪种情况搭配可使试验结果最佳？这些问题正是科技创新的所要解决的问题。而统计正交实验设计法正是一种适于多因素实验的实验设计方法，它虽不能像模拟技术那样地解决间题，但可以在可行域内建立起多种有代表性的实验方案，在此基础上进行模拟实验，并对实验结果进行相应的统计分析，就能够定量地描述可行域空间，从而为模型的优化及找出全域最优解提供依据，同时还可以评价各模拟决策方案的稳定性，以使实验获得理想的结果。本文重点讨论怎样利用正交实验设计法设计模拟实验方案的问题，以及怎样进行实验数据分析的方法。

With the development of science and technology, more and more good test design is requested. Because many research topics are requested to change from the static state(single factor) to the dynamic state (many factors), that is, to study the changing regulation of matter under the situation that many important factors changes in the meantime, how are the results of the test selected and analyzed so as to answer the following questions scientifically: (1) which factor has more influence and which factor has less influence upon the index design? (2) For a certain factor, which data is better? (3) How to match each factor and data to get the best test result? To resolve these problems are exactly wanted in the innovation of science and technology. Coincidently, orthogonal test design is suitable for the test of many factors. Although it can't work like the technique of simulation test, it can establish a variety of representative test scheme with the possible interval. On that basis, an simulation test can be proceeded, and test result can be analyzed statistically accordingly, thus the possible interval can be described in quantity, all of which will provide evidence for excellent model and possible interval. In the meantime, the stability of simulation test scheme are described to acquire an ideal result. This paper mainly discuss how to make use of orthogonal test design to design simulation test scheme and how to analyze test data

1. 姚文俊(2004)。一种基于正交实验设计的遗传算法。中南民族大学学报（自然科学版），2004(1)
2. 华中生(2002)。正交试验设计方法在汽车制动系统试验设计中的应用。运筹与管理，2002(6)
3. 实施科教兴晋正交法在工业生产中大范围推广应用课题组(2003)。正交实验设计法推广应用战略研究。科技情报开发与经济，2003(1)
4. 实施科教兴晋正交法在我省工业企业大范围推广应用课题组(2002)。正交试验设计法在中小企业推广应用的意义。科技情报开发与经济，2002(6)
5. 张永林(1997)。正交实验点的空间分布。数理统计与应用概率，1997(3)
6. 张传平(2001)。中药研究中正交实验设计运用存在的问题探讨。陕西中医学院学报，2001(4)
7. 欧阳明德(1994)。管理系统的正交模拟实验设计研究。华中理工大学学报，1994(9)